Sunteți pe pagina 1din 22

A

REPORT ON

IC-Engine Intake Noise Predictions


Based on Linear Acoustics

BY
RAJESH UPADHYAY
11MEMTMD002

Chapter-1
Introduction:
Air intake noise is created in an internal combustion (IC) engine when the intake valves open to
reveal the pumping motion of the pistons. High amplitude pressure pulsations travel upstream
through the different components in the air intake system, see Fig.1. The sound waves are finally
radiated either through the air intake as orifice noise or through the plastic walls of the intake
system as shell noise due to fluid-structural interaction. The amplitude of these resulting intake
noise sources is strong enough to make them possibly significant contributors to the Pass-bynoise as well as to the sound quality impression of the vehicle. The sound reducing measures
available for the intake system are either relatively space consuming or are reducing the
efficiency of the engine breathing process which is directly linked to fuel consumption and CO2
emissions. In order to meet customer demands of attractive design and reduced vehicle weight
the space dedicated for the powertrain installation is often decreased. New techniques such as
homogeneous charge compression ignition (HCCI), where the engine load can be controlled
without a throttle, might add further complexity to the situation due to increased intake noise
emissions at part load conditions. To meet these requirements, together with intensified noise
legislation and the demand for shorter product development cycle times, better methods are
needed to enable acoustic optimization of the air intake system.
1.2. Modelling intake noise
The gas dynamics of an IC-engine can essentially be described by a set of nonlinear equations
for conservation of mass, momentum and energy [1]. In the general case analytical solutions to
those equations can not be found and numerical models based on various approximations are
necessary. A very powerful and, for IC-engine intake or exhaust ducts, often used simplification
is that of considering one dimensional fields only. This assumption basically implies that the
variables of pressure, density, velocity and temperature are treated as being constant over the
cross-section of the duct under consideration. From this the solution of the coupled non-linear
equations will be greatly simplified. Another convenient simplification is that of assuming small

Perturbations and perform a linearization of the governing equations. When there is a


homogeneous mean flow present the final result will be the convective wave equation and then
also 3D effects can be treated without too much difficulty. If only plane waves are considered,
the wave equation will transfer to a 1D linear wave problem, which can be efficiently analysed
via so called two-port or four-pole methods.
Within the automotive industry the most widely adopted technique for gas exchange studies is to
solve the one dimensional coupled set of non-linear equations using the finite volume or finite
difference method. This technique is used in several commercial softwares e.g., Ricardo/WAVE,
GT-Power and AVL/BOOST, which also provide easy to use graphical user interfaces. The main
purpose of these codes is for tuning of cycle averaged parameters, such as the torque and power
output from engines, but unsteady pressures and flow velocities are additionally provided at
positions distributed throughout the intake and exhaust systems. Hence, they can also be used for
acoustic studies [2]. The boundary condition usually prescribed at any duct orifice is a fixed
pressure corresponding to the ambient conditions. The predicted fluctuating velocities can
thereafter, together with the assumption of spherical or hemispherical radiation, be used to
calculate the noise that is emitted from the intake or exhaust orifice in a post processing step. The
inherent limitation of one dimensional plane wave propagation, however, limits this technique to
sufficiently low frequencies where non-plane wave effects are small. Therefore this type of
model will first fail in large components such as air cleaners. Further limitations,that might not
be important for simulation of engine performance but indeed for acoustics, include difficulties
to apply frequency dependent boundary conditions and include effects of vibrating walls. As a
result the accuracy of the predictions for intake noise is not fully reliable, neither concerning
prediction of absolute sound pressure levels nor resonance frequencies. Several authors have
proposed strategies to improve the predictions of sound based on non-linear 1D gas exchange
simulations. Basically there exist three main groups of methods for how this can be done;
methods that use information from non-linear simulations as input to linear acoustic simulations,
hybrid methods where linear information is inter-changed between the non-linear calculations
and the acoustic simulations and the extension from 1D to a full solution of the 3D non-linear
equations.
A recent example of the first group is the work by Shaw, Moenssen and Montgomery [3], where
fluctuating velocities predicted from 1D gas exchange simulations where used as input to linear

boundary element simulations, but without taking into account the frequency dependence of the
boundary impedance at the coupling section. Another example is the work by Fairbrother, Bodn
and Glav [4] who studied the exhaust noise from a turbo-charged truck engine by using nonlinear 1D computational fluid dynamics (CFD) to extract a linear time invariant source and
thereafter coupled that to linear acoustic two-ports. The predictions of in-duct sound pressure
levels shown were of reasonable agreement to measurements but the free field predictions were
not as good.
A similar exhaust noise study was performed on a four-cylinder naturally aspirated diesel engine
from a passenger car by Hota and Munjal in Ref. [5]. Here, the results at free field appear to be
more accurate; however, only three discrete values of engine speed were reported. Recent
examples of hybrid methods include the work by Payri, Desantes and Torregrosa [6], who also
gives a good review of earlier work, and Chiavola [7]. Earlier work at MWL/KTH dealing with
hybrid methods is described inRefs. [12]-[15]. An example of how non-linear effects can be
included in a one-port model was proposed in Ref. [16] which also includes a literature review of
different source models. The method of coupling 1D to 3D non-linear CFD is provided as a built
in function in some of the commercial softwares [8] but is still not very useful for engineering
noise predictions due to extremely long computational times. An important limitation of linear
acoustic simulations is that they can only be used when the amplitudes of the fluctuating
pressures are small. Often small amplitudes are taken as a relative fluctuation amplitude of 1 %
of the steady state value. This implies that a linear acoustic model is accurate enough up to at
least 154 dB [ref. 210-5] for a steady state value of 1 bar.

Law of Partial Pressures:


Pt = Partial Pressures
Pi = P air + P fuel vap + P water vap

If each is considered as a perfect gas


P=

Mass
M
=
Molecular wt. m

The sound pressure level in the duct just upstream the throttle might exceed this value why nonlinear effects can be of importance, see Fig. 2. The amplitudes close to the inlet valves are
definitely above this limit why the choice of a non-linear technique seems more appropriate in

this region. In principle there are two effects of non-linearites on the wave propagation; wave
steepening (shock forming) and too high local amplitude velocity pulsations e.g. at perforated
elements and narrow constrictions. An important limitation of linear acoustic simulations is that
they can only be used when the amplitudes of the fluctuating pressures are small. Often small
amplitudes are taken as a relative fluctuation amplitude of 1 % of the steady state value. This
implies that a linear acoustic model is accurate enough up to at least 154 dB [ref. 210-5] for a
steady state value of 1 bar. The sound pressure level in the duct just upstream the throttle might
exceed this value why non-linear effects can be of importance, see Fig. 2.
The amplitudes close to the inlet valves are definitely above this limit why the choice of a nonlinear technique seems more appropriate in this region. In principle there are two effects of nonlinearites on the wave propagation; wave steepening (shock forming) and too high local
amplitude velocity pulsations e.g. at perforated elements and narrow constrictions. The
propagation distance xswhere an initial harmonic wave becomes a shock wave is approximately.

1.3 The linear time-invariant one-port source model


A model that can be used to represent an engine as a source must be able to describe the power
input from the source and how incoming waves are reflected by the source. If only plane waves
are considered at the source cross-section the simplest model that can be used is the linear time
invariant one-port source model [10]. The condition of plane waves restricts its validity to
frequencies below the cut-on frequency for the first nonplanar mode.
1.4 Acoustic modelling of special elements
In order to make accurate predictions of intake orifice noise the source as well as the
transmission and the radiation must be described accurately. Examples of interesting components
that are supposed to have large resistive or reactive properties affecting the transmitted sound are
the air cleaner and the charge air cooler. Several parameters can be used to describe the acoustic
performance of a component in an air intake system. These include the transmission loss (TL),
the noise reduction (NR), and the insertion loss (IL). The parameter chosen for comparisons in
Paper B, C and D is the transmission loss which is the difference in sound power level between
the incident and the transmitted sound wave when the test object termination is anechoic. The

standard technique today for measuring acoustic plane wave properties in ducts, such as
absorption coefficient, reflection coefficient and impedance is the two microphone method
(TMM) [17]-[20]. The sound pressure is decomposed into its incident and reflected waves and
the input sound power may then be calculated. Many papers have been devoted to the analysis of
the accuracy of the TMM for example [18]- [20]. Transmission loss can in principle be
determined from measurement of the incident and transmitted power using the two-microphone
method on the upstream and downstream side of the test object, provided that a fully anechoic
termination can be implemented on the outlet side which is difficult in the low frequency region
and with flow. Instead the so-called two-source replacement technique [21] can be used. In this
technique sufficient information for determining the two-port matrix is obtained from two sets of
measurements, one with the source on the upstream side and one with the source on the
downstream side.
1.5 Air cleaner boxes
The component with the largest potential to reduce noise is the air cleaner box, acting as an
expansion chamber, which therefore deserves careful treatment in order to not reduce the
prediction accuracy. Due to geometrical restrictions in the surrounding engine compartment, the
air cleaner has a complicated geometry that normally is not possible to describe analytically.
Predictions in the low frequency region are possible from treating the air cleaner as an expansion
chamber where the area ratio and the length must be correctly given, see e.g. the book by Munjal
[17] for an extensive description. Contribution from higher order modes to the response at higher
frequencies, as is described by Glav in Ref. [22], is not possible to include analytically due to the
non regular shape. More accurate sound transmission predictions can be obtained from numerical
calculations using 3D linear acoustical finite elements or boundary elements, as has been shown
by several authors, see e.g. the work by Herrin et al. [23] for a recent example. However, when
using acoustic finite elements or boundary elements as a tool, the losses in the system are not
predictable. The losses can be caused by several physical mechanisms such as deviations from
adiabatic changes of state, flow and fluid structure interaction. Examples of publications
describing theories for losses due to flow include Boij and Nilsson [24], Alfredsson and Davies
[25], Glav [22] and Allam and bom [26]. Effect on sound transmission in air induction
components due to fluid structural interaction has recently been treated by Marion and Ye [27]

and Alex [28] who stated that the influence of non-rigid walls can be large on the intake orifice.
Both works were, however, based on generic components that where not optimized by means of
wall stiffness. In contrast, air intake system components are normally thoroughly reinforced by
stiffeners in order to reduce the amount of sound that is radiated into the engine compartment.
These reinforcements are also reducing the shift of resonance frequencies that can occur in
resonators with yielding walls. The work in Paper B aims to give information on the importance
of including flow and yielding walls when analysing an air cleaner box that is properly designed
and is taken from a car in series production. The study also includes sound
transmissionmeasurements with and without air filter paper. Surprisingly, as it normally is
neglected in 1D CFD models for predicting engine performance, the filter paper has a large
effect on the transmission loss for frequencies exceeding 500 Hz and must be taken into account
in a complete model. A suggestion for a filter paper model is given by Ih and Kang in Ref. [29]
where a 1D model, originally developed by Allam and bom [26] for diesel particulate filters, is
applied to extract a two-port. At frequencies above 500 Hz, where cross-modes will appear in the
air cleaner and were the filter was shown to be most important, this model will not be sufficient.
1.6 Charge air coolers
Many turbocharged engines are equipped with charge air coolers (CAC), a device used to
increase the overall performance of the engine. The cooling of the charged air results in higher
density and thus volumetric efficiency. Important for petrol engines is also that the knock margin
increases with reduced temperature. The parameters of main interest when designing a CAC are
normally the pressure drop and the heat exchange efficiency. However, what seem to have been
overlooked are the acoustic properties which are still not very well investigated. To the authors
knowledge the sound attenuation properties are only dealt with in two previous publications [30],
[31]. The models in both these references are making use of acoustical two-ports (or four-poles)
to assemble a complete model for a CAC. However, none of them includes a complete treatment
of the losses in the cooling tubes. According to the literature survey in Ref. [30], there are
predictive models available describing the thermal efficiency [32], and also models treating flow
unsteadiness [33]-[35] in CACs. Still they are only evaluated in terms of heat transfer
performance, pressure drop and gas-exchange properties mainly affecting lower frequencies. In
Ref. [36] and [37] Knutsson and bom have presented some initial parts of the work presented

in Paper D in this thesis, which aims to make a complete description of the sound attenuating
properties of a CAC when there is a mean flow present.
One challenge when describing a charge air cooler is to correctly treat the propagation of sound
in narrow cooling tubes with non-circular geometries for different flow speeds. The propagation
of sound in narrow ducts was found by Kirchoff [38] to be dissipative because of viscous and
thermal effects at the duct walls. In this work he formulated the solution to the problem, without
any flow present, as a complicated, complex transcendental equation which so far has not been
solved analytically. In the work by Zwikker and Kosten [39] an approximate solution to the
problem was found from a set of simplified equations.

An extensive overview of other models available for a wide variety of shear wave numbers was
presented by Tijdeman in Ref. [40]. When flow is present the situation is somewhat more
complicated and no complete theory exists. Several authors see Refs. [41]-[46] have, however,
derived solutions based on simplified equations or numerical calculations. For practical
applications the most useful is perhaps the ones proposed by Domumaci in Ref. [41]-[42]. In
Ref. [41] he showed that the equations for sound propagation in a thermo-viscous fluid,
simplified in the manner of Zwikker and Kosten theory [38], could be solved analytically for a
circular pipe with a mean flow profile that is constant over the cross-section. In a later paper [42]
Dokumaci extended the model from [41] to rectangular cross-sections by expanding the solution
in terms of a double Fourier sine series. Other works starting out from essentially the same

equations as used by Dokumaci; include Astley and Cummings [43] and Peat [44], Ih and Park
and Kim [45], and Jeong and Ih [46]. At operating conditions CACs as well as catalytic
converters experience temperature and pressure gradients. The effect of axial axial pressure and
temperature gradients has been treated by Peat [47], Peat and Kirby [48], and Dokumaci [49]. All
the models mentioned above assume laminar flow and do therefore not take into account any
effect of turbulence on the propagation of sound waves. The model by Howe in Ref. [50]
combines the effects of turbulence and viscothermal sub-layers on wave propagations in circular
cross-sections. Although it requires the cross-section to be wide and the turbulent flow to be
fully developed with a constant flow profile, it provides useful information and understanding
concerning the low frequency damping for the cases where the Reynolds number is large. In
Paper D the solution for circular ducts by Dokumaci [41], a modified version of the numerical
solution scheme for arbitrary crosssections derived by Astley and Cummings [43] and the
solution accounting for turbulence in circular ducts by Howe [50] are used to model cooling
tubes in order to find the most accurate solution. Two-ports representing a cooling tube are
extracted from the three solutions, which has not to the authors knowledge previously been done
for the two later. The effect of approximating cross-sections that are shaped as isosceles
trapeziums with circular geometries, where the hydraulic diameter is equivalent, is studied for
cases where laminar or turbulent flow is present. Another challenge concerning charge air cooler
modelling concerns coupling of several ducts into one volume, where higher order modes are
present, in order to extend the frequency range. The commercial FE software LMS Sysnoise
provides the ability to couple two arbitrary volumes together via two-ports [51] and could be
used for this purpose. This procedure requires that the complete model is solved for each
frequency for each load case; for models with high element resolution time consuming as well as
expensive. A good method to shorten iteration time, if the properties of the air mixture and the
geometry of the tanks are fixed, is to use the multi-port approach that is
described in Paper D. The algebra for the multi-ports, based on the admittance relationships
between the ports, is derived and presented in an attractive form for easy implementation.

CHAPTER-2
Review Literature:
2.1. Paper A: IC-engine acoustic source data from non-linear simulations
Non-linear one-dimensional (1D) CFD time domain prediction codes are used to calculate the
performance of the gas exchange process for IC-engines. These softwares give time-varying
pressures and velocities in the exhaust and intake systems. They could therefore in principle be
used to predict radiated orifice noise. However, the accuracy is not sufficient for them to be used
as a virtual design tool. More accurate results might be provided by dividing the problem into a
source domain and a transmission domain and use linear 3D frequency domain codes to describe
the transmission part. Radiated shell noise and frequency dependent damping could also be
included in the frequency domain models. The simplest source model for simulation of the
dominating engine harmonics is the linear time invariant 1-port model. The 1-port source data is
usually obtained from experimental tests where the multi-load methods and especially the twoload method are most commonly used. The main limitations of these tests are that they are time
consuming, expensive and require the engine, which prevents them from being used for early
predictions. It would therefore be of interest to extract the acoustic source data from the existing
1D CFD gas exchange models. This paper presents a comparison between 1-port source data,
obtained by measurements on a six-cylinder personal car petrol engine, and by 1D simulations of
identical intake systems on the same engine. The degree of non-linearity in the results is
discussed as well as the choice of source type and its relation to engine properties. The results
show that it is possible to obtain reasonably accurate source strength as well as source impedance
estimates, for the intake side, from 1D gas exchange simulations.
2.2. Experimental investigation of the acoustic effect of non-rigidwalls in IC-engine intake
systems
This paper presents the results of an experimental study of the acoustic properties of an
automotive air intake system. Modern air intake systems for personal cars are largely made of
plastic materials. When trying to make an acoustic model for the sound transmission through the
intake system it is questionable if the walls of different components can be modelled as

acoustically rigid. The acoustic losses which determine the transmitted sound level at breakthrough frequencies may be especially difficult to model. To find out if acoustic losses associated
with wall vibrations or with flow interaction are dominating for a typical intake system an
experimental investigation has been performed. The acoustic plane-wave two-port matrices have
been measured, using the two-source replacement technique, for the complete intake system and
for separate parts. Measurements were made in a flow test rig without flow and for two different
flow speeds. The measurements were then repeated with sand loading on the walls to reduce wall
vibrations. Finally the measurements were repeated with the filter paper
removed. The results indicate that flow related losses dominates at low frequencies while losses
associated with acoustically non-rigid walls are more important at higher frequencies, giving an
additional transmission loss in the order of 2 dB for the complete system. However, most
important at higher frequencies is the filter paper that can add more than 10 dB at frequencies
where the reactive effects in the system are small. All three effects will have to be taken into
account in an accurate model. Comparisons are also made between results from an acoustic FEmodel and experimental results.
2.3. Paper C: Prediction of IC-engine intake orifice noise using 3Dacoustic modelling and
linear source data based on non-linear CFD.
Shorter product development cycles, densely packed engine compartments and intensified noise
legislation increases the need for accurate predictions of engine air inlet noise at early stages.
Non-linear one dimensional (1D) computational fluid dynamic (CFD) time domain prediction
codes are used for calculation of the performance of the gas exchange process of internal
combustion (IC) engines. These codes give time varying pressures and velocities in the exhaust
and intake system. From using the calculated velocity fluctuations at the air intake and the
assumption of monopole radiation the noise emanating from the orifice can be predicted in a
linear post processing step. However, the accuracy is unreliable for intake systems exhibiting a
three dimensional nature. The use of linear three dimensional acoustic frequency domain
methodologies such as finite elements for the transmission might provide more accurate results.
In the present study linear acoustic one-port source data for a six cylinder naturally aspirated
engine is extracted from 1D gas exchange analysis. The sound transmission from the source

reference cross-section and through the air cleaner is represented by one single two-port that is
extracted from acoustic 3D finite element analysis as well as from experiments. The radiated
intake orifice noise is obtained from combining the source data and the two-port with the
terminating inlet duct (the dirty air duct), represented by boundary elements. Finally the
complete coupled model is validated using intake orifice noise measurements obtained in an
engine test bench. The results are as good as, or for the case of higher frequencies even better,
than the predictions from using non-linear 1D CFD for the complete system together with a
linear monopole radiation model.
2.4 Sound propagation in narrow channels including effects ofviscothermal and turbulent
damping with application to charge air coolers
Charge air coolers (CACs) are used on turbo charged internal combustion engines to enhance the
overall gas exchange performance. The cooling of the charged air results in higher density and
thus volumetric efficiency. Important for petrol engines is also that the knock margin increases
with reduced charge air temperature. A property that is still not very well investigated is the
sound transmission through a CAC. The losses, due to viscous and thermal boundary layers as
well as turbulence, in the narrow cooling tubes results in frequency dependent attenuation of the
transmitted sound that is significant and dependent on the flow conditions. Normally, the crosssections of the cooling tubes are neither circular nor rectangular, why no analytical solution
accounting for a superimposed mean flow exists. The cross-dimensions of the connecting tanks,
located on each side of the cooling tubes, are large compared to the diameters of the inlet and
outlet ducts. Three dimensional effects will therefore be important at frequencies significantly
below the cut-on frequencies of the inlet/outlet ducts. In this study the two dimensional finite
element solution scheme for sound propagation in narrow tubes, including the effect of viscous
and thermal boundary layers, originally derived by Astley and Cummings [43] is used to extract
two-ports to represent the cooling tubes.
The approximate solutions for sound propagation, accounting for viscothermal and turbulent
boundary layers derived by Dokumaci [41] and Howe [50], are additionally calculated for
corresponding circular cross-sections for comparison and discussion. The two-ports are thereafter
combined with numerically obtained multi-ports, representing the connecting tanks, in order to

obtain the transmission properties for the charged air when passing the complete CAC. An
attractive formalism for representation of the multi-ports based on the admittance relationship
between the ports is presented. From this the first linear frequency domain model for CACs,
which includes a complete treatment of losses in the cooling tubes and 3D effects in the
connecting tanks is extracted in the form of a two-port. The frequency dependent transmission
loss is calculated and compared to corresponding experimental data with good agreement.

CHAPTER-3
System description
An intake system for a naturally aspirated multi-cylinder IC-engine is mainly composed of an
intake manifold, a throttle, an air cleaner, ducts and occasionally resonators. The primary task of
the system is to provide the engine with clean air at appropriate temperature with as small
pressure drop as possible. Among the secondary tasks, the ability to reduce sound is crucial since
the orifice intake noise is one of the main contributors to the Pass-by noise legislation for
passenger cars. The dominating part of the intake noise from a naturally aspirated engine is
originating from the movement of the pistons and valves and travels as plane waves upstream the
ducts to the intake orifice, where it is radiated to the surroundings. The component with the
largest potential to reduce noise is the air cleaner box acting as an expansion chamber. Due to
geometrical restrictions in the surrounding engine compartment, the air cleaner has a
complicated geometry that normally is not possible to describe analytically. Predictions are
possible either from measurements or by numerical calculations. However using acoustic finite
elements as a tool, the losses in the system are not predictable. The losses can be caused by
several physical mechanisms such as deviations from adiabatic changes of state, flow and fluid
structure interaction. This project aims to study the losses due to fluid structure interaction and
flow for a typical air cleaner system.
Several parameters can be used to describe the acoustic performance of a duct element such as an
air intake system. These include the transmission loss (TL), the noise reduction (NR), and the
insertion loss (IL). The parameter chosen for comparisons in this investigation is the
transmission loss which is the difference in sound power level between the incident and the
transmitted sound wave when the test object termination is anechoic. The standard technique
today for measuring acoustic plane wave properties in ducts, such as absorption coefficient,
reflection coefficient and impedance is the two-microphone method (TMM) [2-5,9]. The sound
pressure is decomposed into its incident and reflected waves and the input sound power may then
be calculated. Many papers have been devoted to the analysis of the accuracy of the TMM for
example [2-4]. Transmission loss can in principle be determined from measurement of the
incident and transmitted power using the two-microphone method on the upstream and
downstream side of the test object provided that a fully anechoic termination can be implemented

on the outlet side, which is very difficult in the low frequency region and with flow. Instead the
so-called two-source replacement technique [8] has been used. In this technique sufficient
information for determining the two-port matrix is obtained from two sets of measurements, one
with the source on the upstream side and one with the source on the downstream side [7].

Shorter product development cycles, densely packed engine compartments and intensified noise
legislation increase the need for accurate predictions of engine air inlet noise at early stages.
Non-linear one dimensional (1D) computational fluid dynamic (CFD) time domain prediction
codes are used for calculation of the performance of the gas exchange process of internal
combustion (IC) engines. These codes give time varying pressures and velocities in the exhaust
and intake system. From using the calculated velocity fluctuations at the air intake orifice and the
assumption of monopole radiation the noise emanating from the orifice can be predicted in a
linear post processing step. However, the accuracy is unreliable for intake systems exhibiting a
three dimensional nature. The use of linear three dimensional acoustic frequency domain
methodologies such as finite elements for the transmission might provide more accurate results.
In the present study linear acoustic one-port source data for a six cylinder naturally aspirated
engine is extracted from 1D gas exchange analysis. The sound transmission from the source
reference cross-section and through the air cleaner is represented by one single two-port that is
extracted from acoustic 3D finite element analysis as well as from experiments. The radiated

intake orifice noise is obtained from combining the source data and the two-port with the
terminating inlet duct (the dirty air duct), represented by boundary elements. Finally the
complete coupled model is validated using intake orifice noise measurements obtained in an
engine test bench.

Inlet Pressures and Densities


P air
P air
=
Pinlet P air + P fuel + P water
M a / ma
=
M a / ma + M f / m f + M w / m w

ma = 29

mw = 18

mgas = 113

Fc = chemically correct mix


Fi = % vaporized (Fc)
The tanks of the chare air cooler used in this study are very similar in size and shape. The
volume of the inlet tank is 853 cm3 and that of the outlet tank is 896 cm3. There is a circular
opening in the wall of the outlet tank designed to support a temperature transducer. In all
measurements and calculation results presented here this hole is carefully plugged to create a
smooth wall. In order to get a fast and simple meshing procedure, linear tetrahedral elements
with four nodes are used to model the volumes and linear wedge element with six nodes for the
pipes. The use of hexahedral elements with eight nodes would certainly be more efficient but the
process of creating the element mesh is still not possible to be fully automated and hence very
cumbersome.
Parabolic tetrahedral elements would of course also be an efficient choice but for this case it was
decided to use the linear element with very high resolution. The chosen element size is 5 mm
which yields more than 40 elements per wavelength for analyses performed at a frequency of
1600 Hz at cold conditions and more than 50 for thewarmer air present at full load and medium
engine speed. The total number of nodes is about 23 000 in each tank, built up using 76 000
elements in the inlet tank and 83 000 in the outlet tank. This finite element mesh is shown in Fig.
5.

Figure 5. Finite element mesh of inlet and outlet tanks (the tanks have been moved together in
the image).
The circular pipes, that connect the tanks to the rest of the gas exchange ducting of the engine,
have diameters equal to 60 mm. Short parts of the measurement ducts, with the diameter 66 mm,
are also included in the model in order to make sure that the waves are plane after the 90 degree
bend in each tank. The same technique is also used at the connection to the cooling tubes where
acoustical near fields are present due to the sudden contraction and expansion. When using
acoustical finite elements it is well known that the numerical integration of the pressure at a
boundary is done exactly. However, the acoustic velocityboundary condition is obtained only in a
weak sense since it is of Neumann type. The prediction of the acoustic velocity at a boundary is
obtained from extrapolation of the results at the integration points inside the elements to the
boundary. To account for the inaccuracy resulting from the extrapolation the elements at all
openings are made extremely short (0.1 mm).
At the outlet from each cooling tube there is an abrupt expansion, see Fig. 7a, which needs a
separate two-port for the cases when there is a mean flow present. Flow separation will be
present hence entropy is generated due to the dissipation in the turbulent mixing region
downstream the flow separation. This interaction between the acoustic field and the flow
separation is, however, very complex and complete analytical models are complicated. An
example of such a model is the work by Boij and Nilsson [36]. Several authors have suggested
models based on simplified velocity fields. The model used here is based on the simple analysis
in Ref [37] which has been found to give good results for low Mach numbers and rather low

frequencies [38] - [40]. An area contraction is formed at the inlet of the cooling tubes. Depending
on the flow situation, a vena contracta can be formed just after the area contraction.

A complete CAC linear frequency domain hybrid-model based on coupling multi-ports,


representing the inlet/outlet tanks, to resistive two-ports, representing the cooling tubes, is
presented. The model is validated for an air-to-air charge air cooler used in passenger cars, see
Fig. 3. The cooling tubes in this CAC are equipped with turbulators, as shown in Fig. 4, a folded
metal sheet with large area, used to improve the heat exchange efficiency of the cooling tubes.
This installation creates narrow, almost triangularly shaped (or isosceles trapezium), internal
axial channels with dimensions equivalent to a hydraulic diameter between 2.5 and 3 mm. The
propagation of sound in such tiny ducts was found by Kirchhoff [11] to be dissipative because of
viscous and thermal effects at the pipe walls. In this work he obtained the solution to the
problem, without any flow present, as a complicated, complex transcendental equation which so
far has not been solved analytically. In the work by Zwikker and Kosten [12] an approximate
solution to the problem was found from a set of simplified equations. When flow is present the
situation is somewhat more complicated and no complete theory exists. Several authors [13]-[22]
have, however, derived solutions based on simplified equations or numerical calculations.

FUTURE SCOPE:
Interesting future research concerning source data extraction includes more investigations on
how to choose acoustic loads for the multi-load method. Using 1D CFD opens the opportunity of
optimization of this process in order to further increase the accuracy of the estimated source data.
Furthermore, it would be interesting to verify the linear time-invariant source model on a fourcylinder engine, were the reflections from the expansion at the air cleaner will affect the
performance of the engine in larger extent than on a six-cylinder engine. The task of developing
an orthotropic model for air cleaner paper is the most important in order to improve simulation
results, for sound propagation through air intake systems on naturally aspirated engines, above
500 Hz. An interesting reference is the work by Ih and Kang [29] who used a 1D model to
simulate wave propagation through a folded filter paper. In order to extend the accuracy of this
model to exceed 500 Hz three dimensional effects will have to be taken into account. The
vibration of the filter paper might also influence the sound transmission and should therefore also
be included in a complete model. The charge air cooler measurements and simulations in Paper
D expose the need for a wave propagation model that fully includes the effect of turbulence
damping for low frequencies.

REFERENCES:
[1] D. E. Winterbone, R. J. Pearson, Theory of engine manifold design, Wave action methods
for IC engines, Professional engineering publishing, London, 2000.
[2] WAVE, Acoustic Manual, 2001, Ricardo.
[3] C. E. Shaw, D. J. Moenssen, W. C. Montgomery, On the use of BEA with engine simulation
as an input to predict air induction inlet noise, SAE Technical Paper 2005-01-2350, 2005.
[4] R. Fairbrother, H. Bodn, R, Glav, Linear acoustic exhaust system simulation using source
data from non linear simulation, SAE Technical Paper 2005-01-2358, 2005.
[5] R. N. Hota, M. L. Munjal, Hybrid approach to predict exhaust noise of naturally aspirated
engine with commercial mufflers, SAE Technical Paper 2007-26-039, 2007.
[6] F. Payri, J. M. Desantes, A. J. Torregrosa, Acoustical boundary conditions for unsteady
one-dimensional flow calculations, Journal of Sound and Vibration 188 (1) (1995) 85110.
[7] O. Chiavola, Multi-dimensional CFD-transmission modelling of IC engine intake and
exhaust systems, Journal of Sound and Vibration 256 (2) (2002), 835-848.
[8] GT-Power, Users Manual and Tutorial, GT-suite version 6.1, 2004, Gamma Technologies.
[9] A. D. Pierce, Acoustics - An introduction to its physical principles, The Acoustical Society
of America, New York, 1989.
[10] H. Bodn, M. bom, Modelling of fluid machines as sources of sound in duct and pipe
system, ActaAcustica3 (1995), 549-560.

[11] H. Bodn, Characterisation of fluid-borne sound sources and structure borne sound
sources, Proceedings of the 9th International Congress on Sound and Vibration, Orlando,
USA, 2002.
[12] F. Albertson, Negative resistances in nonlinear lumped acoustic systems and IC-engine
source data, Forum Acusticum99, 1999.
[13] F. Albertson, H. Bodn, Methods for prediction of the sound generation from the ICengine
exhaust. Proceedings of the 6th International Congress on Sound and Vibration,
Copenhagen Denmark, 1999.
[14] H. Bodn, F. Albertson, Application of the multiple load method for non-linear sources.
Proceedings of the 7th International Congress on Sound and Vibration, GarmischPartenkirchen, Germany, 2000.
[15] F. Albertson, J. Gilbert, Harmonic balance method used for calculating the steady state
oscillations of a simple one-cylinder cold engine, Journal of Sound and Vibration 214 (4)
(2001), 541-565.
[16] H. Rmmal, Methods for modelling and characterization of in-duct sources, Lic. thesis,
The Royal Institute of Technology, Stockholm, Sweden, 2005.
[17] M. L. Munjal, Acoustics of ducts and mufflers, Wiley, New York, 1987.
[18] J. Y. Chung, D. A. Blaser, Transfer function method of measuring in-duct acoustic
properties I. Theory, Journal of the Acoustical Society of America, 68 (1980), 907-913.
[19] H. Bodn, M. bom, Influence of Errors on the Two-Microphone Method for Measuring
Acoustic Properties in Ducts, Journal of the Acoustical Society of America, 79 (2) (1986),
541-549.

[20] H. Bodn, M. bom, Error analysis of two-microphone method measurements in ducts


with flow, Journal of the Acoustical Society of America, 83 (2) (1988), 2429-2438.
[21] M. bom, Studies of sound propagation in ducts and mufflers, PhD thesis, The Royal
Institute of Technology, Stockholm, Sweden, 1989.
[22] R. Glav, On acoustic modelling of silencers, Ph. D. Thesis, The Royal Institute of
Technology, Stockholm, Sweden, 1994.
[23] D. W. Herrin, Z. Tao, J. Liu, A. F. Seybert, Using boundary element analysis to analyze
multi-component exhaust systems, SAE Technical Paper 2007-01-2182, 2007.
[24] S. Boij, B. Nilsson, Scattering and absorption of sound at flow duct expansions, Journal of
Sound and Vibration 289 (3) (2006), 577-594.
[25] R. J. Alfredson, P.O.A.L. Davies, Performance of exhaust silencer components, Journal of
Sound and Vibration 15 (2) (1971) 175-196.

S-ar putea să vă placă și